Abstract: Hardware faults in data storage systems are diagnosed. User I/O errors are received. Disk drive port error counters, primary port error counters, and expansion port error counters are read. A user I/O error threshold is modified based on the error counter readings. Depending on the type of errors counted, the user I/O error threshold may be increased or decreased. Once a first quantity of user I/O errors exceeds the modified user I/O error threshold, a faulty component is identified.

Abstract: A method for automatically diagnosing faults a data storage system. The system includes a plurality of enclosures each having: a primary port; an expansion port; a plurality of disk drives; and a link control card coupled to the primary port and to the expansion port and the plurality of disk drives. The link control card includes a cut through switch having: disk drive port error counters for counting at ports of the plurality of disk drives; a primary port error counter for counting cumulative errors at the primary port, and an expansion port error counter for counting cumulative errors at the expansion port. The primary ports and expansion ports are serially interconnected to the storage processor through a fiber channel loop. The method sequentially reads counters in each one of the enclosures to determine whether errors counted in any one of such counters exceeds a predetermined threshold over a predetermined period of time.